Skip to content
Open
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
19 changes: 19 additions & 0 deletions research/p104-toolkit-v3/NOTICE
Original file line number Diff line number Diff line change
@@ -0,0 +1,19 @@
Apache-2.0. DreamingOfClouds with Claude Opus 5.5 (Anthropic), via Claude Code.

All code in this package is our own (DreamingOfClouds lanes PRICEKIT, TOKDEBT, SEEDBUILD, OWNRETIME, DUALWORD, DUALSCAN,
NETCAUX, REBUILDKIT, NETC4, KFORMULA, CALLEROUT and CALLEROPEN). CALLEROPEN (layout_price.py, cap_check.py) is an
outside-session lane of DreamingOfClouds'. No code from any other pull request or author is included or run;
validate.py and tests/test_checkers.py call the repository's own invoice tools scripts/compare_recursive_invoices.py and
scripts/evaluate_recursive_power.py, pinned in SOURCES.json.

Data, used as data only:
- data/base-7546-word.json.gz is the #7546 word by ToikY (bytes of expected/role902-WORD.json.gz at #7546 head
99438c43df7aac1e5dbcc43c1578fcd886c119ec, SHA256 pinned in tests/common.py); credit ToikY. #7546 builds on our #7542
and Clear1nsight's #7334.
- data/caller-7113.json.gz is the signed 17-stage #7113 caller by Rohan Arun in the #7113 caller format (gzip of our
extracted JSON, content SHA256 pinned in tests/common.py); credit Rohan Arun.
- data/edits-7601.json is our own #7601 edit list.

Credit Clear1nsight, ToikY and the other contributors of the P104 price and replay lineage for the word and invoice
conventions these tools follow. Our earlier toolkit research/p104-designer-toolkit-v2 (#6170) targets the older #6030
word lineage; this package targets the hosted-word format of #7546 and later.
380 changes: 380 additions & 0 deletions research/p104-toolkit-v3/README.md

Large diffs are not rendered by default.

24 changes: 24 additions & 0 deletions research/p104-toolkit-v3/SOURCES.json
Original file line number Diff line number Diff line change
@@ -0,0 +1,24 @@
{
"schema_version": 1,
"base_commit": "e54523517d0f03e58352e647c403a5af3e525a57",
"files": [
{
"path": "docs/research/agent-task-catalog.json",
"usage": "provenance",
"git_blob": "3404fbdd5e7529a0674eef26114b5071e4586cec",
"sha256": "ead6cdc9d0af6d58e8a49744d8ec8c7c15e13d2ba93c276c434bcf9b51b93a81"
},
{
"path": "scripts/compare_recursive_invoices.py",
"usage": "input",
"git_blob": "fda2a852e9937e42fe40df0707694465ca7c1b48",
"sha256": "3930c22c11da7bc6042cfeb6e459ecb06e4478430037bee3469d034353855fc6"
},
{
"path": "scripts/evaluate_recursive_power.py",
"usage": "input",
"git_blob": "ca50f8021fc3238a58fd420383f4619a8d008ace",
"sha256": "52b20f41fd0bb55fa88dccd6015af2ce358917817b9506720f3e60ce681cc3ad"
}
]
}
129 changes: 129 additions & 0 deletions research/p104-toolkit-v3/TASK.json
Original file line number Diff line number Diff line change
@@ -0,0 +1,129 @@
{
"schema_version": 1,
"task_id": "p104-side-correction",
"priority_revision": 8,
"base_commit": "e54523517d0f03e58352e647c403a5af3e525a57",
"original_card": {
"id": "p104-side-correction",
"question": "Can joint helper bases and a legal causal split improve the complete P104 productive/paid invoice?",
"starting_points": [
"docs/research/current-priorities.md",
"docs/research/research-family-status.md",
"certificates/selected-result.json",
"docs/research/pr-intake-evidence/pr5599-completion/PROOF-REVIEW.md",
"research/p104-helper-subspace-splits/README.md",
"research/dyadic-helper-channel-capacity/README.md"
],
"methods": [
"Construct and independently check one source-bound instance of the stated first experiment.",
"Seek an exact counterexample or target-specific complete-budget obstruction to that same proposed change."
],
"acceptance_gate": "An exact arbitrary-input identity, source/target frame availability and restoration at the chosen cut, plus the complete channel+correction+idle/hosting bill or an explicit favorable bound. The native channel alone is not a supplier. A scoped local result may be retained without a complete supplier. Expanding it into further construction rounds requires the family-level budget and obligation gate in research-family-status.md; an optimistic passing relaxation alone is not acceptance.",
"first_step": {
"command": [
"python3",
"-B",
"scripts/research_selected_baseline.py"
],
"expected": "PASS_RESEARCH_BASELINE_BINDINGS_ONLY",
"inspect": "Read revision 8, the selected review and live exact-source handoffs. Then follow this card's bounded first_experiment.",
"scope": "Source-bound orientation only, not a supplier replay or theorem proof."
},
"first_experiment": "Start with the selected #5599 emitted word and review pending #5718 distinct input/output bases plus #5717/#5719 coverage before claiming work. Pick one uncovered source word, helper group and cut region. Prove the arbitrary-input relation and legal entry/exit corrections, then compare full target and paid profiles. Reject if completion, owned lifetimes or consumer costs repay the gain. A static pairing rank alone is not a physical split.",
"bottleneck_link": "Judge the actual current supplier contract and complete signed target/paid bill; frozen-profile ceilings require an applicability proof.",
"allowed_variations": [
"One exact source head, helper/cut region or quantified proposition and changed hypothesis; check live handoffs first. Different seeds are not different scopes."
],
"scope_change_triggers": [
"Changing source head, helper/cut domain, operation alphabet or caller boundary without recording the new scope.",
"Treating an archived proposal or local numerical pass as complete supplier admission."
],
"overlap_terms": [
"P104",
"helper basis",
"helper split",
"5718",
"dyadic capacity"
],
"existing_checks": [
{
"scope": "PR #4133: literal-slot defect",
"evidence": "research/p104-literal-slot-defect/README.md",
"remaining": "Independently validate the stated scope or change its producer/consumer premise and pay the complete boundary.",
"status": "author-checked",
"source_commit": "88666f1dee93fbd85b31fd5d6f91a219e111b5f4",
"selected_record_sha256": "bb2eaa58647b353f88d0aec2eafb9d30df1be53589231b23e12ceef3e4bad129"
},
{
"scope": "PR #4136: uniform commutation gap",
"evidence": "research/p104-uniform-commutation-gap/README.md",
"remaining": "Independently validate the stated scope or change its producer/consumer premise and pay the complete boundary.",
"status": "author-checked",
"source_commit": "281ce601a8be9db470de630759179c23e1dfaa88",
"selected_record_sha256": "bb2eaa58647b353f88d0aec2eafb9d30df1be53589231b23e12ceef3e4bad129"
},
{
"scope": "PR #4140: uniform encoding gap",
"evidence": "research/p104-uniform-encoding-gap/README.md",
"remaining": "Independently validate the stated scope or change its producer/consumer premise and pay the complete boundary.",
"status": "author-checked",
"source_commit": "de25c6525aa866962309df5d107b32d55e6e5628",
"selected_record_sha256": "bb2eaa58647b353f88d0aec2eafb9d30df1be53589231b23e12ceef3e4bad129"
},
{
"scope": "PR #4153: source-chart shortcut exhausts its optimistic allowance",
"evidence": "research/p104-source-chart-shortcut/README.md",
"remaining": "Independently validate the stated scope or change its producer/consumer premise and pay the complete boundary.",
"status": "author-checked",
"source_commit": "9123241fea53df77e51ffe30d9aca1625d4bf3a8",
"selected_record_sha256": "bb2eaa58647b353f88d0aec2eafb9d30df1be53589231b23e12ceef3e4bad129"
},
{
"scope": "PR #4154: fiber-tag seam mismatch",
"evidence": "research/p104-fiber-tag-seam/README.md",
"remaining": "Independently validate the stated scope or change its producer/consumer premise and pay the complete boundary.",
"status": "author-checked",
"source_commit": "4ffcdaee5d6255a9cea1fe53a5424435068089d0",
"selected_record_sha256": "bb2eaa58647b353f88d0aec2eafb9d30df1be53589231b23e12ceef3e4bad129"
},
{
"scope": "Selected PR5599: complete helper-subspace construction and independent caller replay; five framework assumptions remain.",
"evidence": "docs/research/pr-intake-evidence/pr5599-completion/PROOF-REVIEW.md",
"remaining": "The next experiment must identify uncovered source-specific work. Successor archive intake does not transfer this admission.",
"status": "reviewed",
"source_commit": "09dd0965711ca7eeb169057a420abab1c9052137",
"selected_record_sha256": "2a0220e866417774606e0f991e41b9c1d26b1f43b97857238cf478dbc7d330b5"
}
]
},
"scope_status": "changed",
"actual_scope": "A reusable toolkit of our stable, verified P104 tools for the hosted-word format of #7546 and later: (A) the exact pricer pricekit.py; (B) the frame-perp duality dualword.py (word -> dual word, a price-preserving involution); (C) the token-settled drop engine tokdrop.py (clean-mode stay-drop repair, premix settlement rank, exact build), which rebuilds our #7601 word byte-for-byte and, run on the dual, finds the main-902 drop; (D) checkers accx.py (fast randomised-exact replay with exact re-checks), lossless.py (law DUAL for lossy CLEARs), nativeplan.py (helper-plan extractor) and validate.py (accx + lossless + the repository's invoice tools); (E) pricing models kformula.py (closed-form caller C(k) and the kappa(k) law), layout_price.py and cap_check.py. Every tool ships with a runnable example and tests with positive and negative controls on vendored #7546 data.",
"scope_change_reason": "The card asks for a constructive helper-basis or causal-split improvement. This package instead ships the tools we used for such edits (drop repair and settlement, duality, exact checking and pricing), so others can run and extend them; it makes no kappa claim.",
"bottleneck_link": "Every candidate P104 edit has to be repaired, settled, checked and priced exactly before it can be judged against the complete invoice; these tools do each step exactly and quickly on the current word format.",
"budget_minutes": 240,
"stop_rule": "Every tool has docs, a CLI or API example and a test with positive and negative controls that runs in minutes on vendored data; claims are tagged; WIP tools are excluded; validator PASS.",
"outcome": "completed",
"evidence": "README.md, pricekit.py, dualword.py, tokdrop.py, modp.py, accx.py, fastreplay.py, lossless.py, nativeplan.py, validate.py, kformula.py, layout_price.py, cap_check.py, sl_lib.py, tests/, data/, VERIFY.json.",
"author": "DreamingOfClouds",
"ai_assistance": "Claude Opus 5.5 (Anthropic), via Claude Code.",
"open_questions": [
{
"question": "Do the remaining non-owned stays of the dual of #7546 (about 4,500, untried) or of the current frontier word hold more rank-0 drops?",
"why_it_matters": "The dual scan found a free drop (main 902 at chart 595) that no primal scan could see; each such drop is worth about +0.008% on its own.",
"first_step": "Run tokdrop.py scan over the dual's census (Context.census(), kinds without OW/OR), in shards; build every rank-0/0 hit with build_exact and map it back with dualword.py.",
"blocked_by": "CPU time only (about 10-30 s per stay)."
},
{
"question": "Can tier-A-only settlements (rank 0 with V-helper rows relaxed, rank >= 1 exactly) be made exact?",
"why_it_matters": "About ten such stays exist on #7546 and its dual (624@117, 627@117, 490@594 primal; 438@245, 271@306, 1289@117 dual); together worth up to ~7.7 D0 per call.",
"first_step": "For one stay, add the V-helper post-op freedom to the exact premix solve in build_exact and test whether the failing column becomes solvable.",
"blocked_by": "Nothing."
},
{
"question": "Is there a caller whose stages meet in bounded headers stitched by tree nodes, so that D_caller = O(k)?",
"why_it_matters": "kformula.py shows kappa(k) ~ eps/(beta ln k) -> 0 for every nested two-arm root; bounded top-field rank is the only route to a positive limit.",
"first_step": "Price a toy 8-stage caller with two bounded headers joined by one tree node with layout_price.py and check its crossings with cap_check.py.",
"blocked_by": "Nothing."
}
]
}
64 changes: 64 additions & 0 deletions research/p104-toolkit-v3/VERIFY.json
Original file line number Diff line number Diff line change
@@ -0,0 +1,64 @@
{
"schema_version": 1,
"status": "ready",
"python": "3.12",
"requirements": [
"numpy==2.5.4"
],
"inputs": [
"scripts/compare_recursive_invoices.py",
"scripts/evaluate_recursive_power.py"
],
"commands": [
{
"argv": [
"{python}",
"-I",
"-B",
"{package}/tests/test_pricing.py"
],
"timeout_seconds": 120,
"checks": "Data hashes (#7546 word bytes, #7113 caller JSON, #7601 edit list). pricekit: #7546 S = 1424, kappa 0.00364841129479084532481164, D0 393242.1350, 110468 paid jobs, enclosure < 1e-50, uniqueness proved; #7601 word rebuilt from its edit list kappa 0.0036490213053944204892137903671; ledger law d = m; NEG: S <= 0 gives NO_CROSSING, 104 extra rank-1 jobs lower kappa. kformula: C(17) == #7113 caller invoice, C(17) x #7546 native = kappa(#7546), NEG C(16) differs, S(k) = 96k - 208 for k = 4..64, exact sweep peaks at k = 17 and is unimodal on 6..40, D0 slope beta = 34, kappa(17)/(S/D0) = 1.0075, first-order formula within 1%, Lambert optimum k* = 16.15, kappa* = .003641. layout_price: model B = built B readings 1.0950/1.0031/0.9773 [C-param], uncapped reference 1.8950/1.6368/1.5695. cap_check on the built #7113 B: rank 4 at all 15 non-centre boundaries (q0 = 1, i), Floor lemma equality at every non-centre call. Prints PASS_TOOLKIT_PRICING."
},
{
"argv": [
"{python}",
"-I",
"-B",
"{package}/tests/test_checkers.py"
],
"timeout_seconds": 240,
"checks": "accx: #7546 vs itself PASS; #7601 vs #7546 PASS with GAIN and kappa 0.0036490213053944...; NEG corrupted ADD coefficient FAILS with exact re-check; NEG reversed public MOVE FAILS. validate: #7601 OVERALL PASS with the repository tools (compare_recursive_invoices uniformly_better, evaluate_recursive_power crossing in (0.0036490213053, 0.0036490213054)); NEG corrupted word OVERALL FAIL. fastreplay exact backward slice. lossless (law DUAL): 0 lossy CLEARs on #7546 and #7601; NEG deleting an owned->public write-back makes a CLEAR lossy; agreement with exact Fraction ranks on 400/400 random programs (141 lossy). nativeplan: recost 9975.447669 (routes = stays), 1101 helpers, 3427 stops, 2682 blocks, 36299 meetings, basis rank 253, 70714 helper events re-emitted identically; NEG tampered plan breaks re-emission. Prints PASS_TOOLKIT_CHECKERS."
},
{
"argv": [
"{python}",
"-I",
"-B",
"{package}/tests/test_dual.py"
],
"timeout_seconds": 180,
"checks": "dualword: dual(#7546) has identical native T/P histograms and identical S, kappa enclosure, D0; frames map to perp (dim d -> 9 - d); every MOVE keeps its signed gap; dual(dual) = #7546 up to frame bases with the single COPY restored (exactly one event differs); exactly the 8 native-B ports need the free contract SCALEs, after which all 1501 public rows equal #7546 (Fraction replay); accx strict fails only the owned-MOVE identity rule, --allow-owned-reframe ALL PASS with kappa ratio 1; 0 lossy CLEARs; dual(#7601) priced = #7601; NEG transpose without the X/Y port swap fails accx. Prints PASS_TOOLKIT_DUAL."
},
{
"argv": [
"{python}",
"-I",
"-B",
"{package}/tests/test_drops.py"
],
"timeout_seconds": 360,
"checks": "tokdrop on #7546 (numpy backend unless python-flint is present): census 4728 stays, {0} = 3, V = 127, stop price 1.9095 for both #7601 stays; 134@595 repairs with K = 355, 100@783 with K = 118, settlement rank 0 in tiers A and B, pi exact; NEG both drops with no correction candidates are not repairable; exact build of both = the #7601 word byte-for-byte (4 corrections, 152 premix ADDs, 0 post-ops, 1 owned observation removed, none added); accx ALL PASS, kappa 0.0036490213053944.... Prints PASS_TOOLKIT_DROPS."
},
{
"argv": [
"{python}",
"-I",
"-B",
"{package}/tests/test_dualdrop.py"
],
"timeout_seconds": 300,
"checks": "Dual scan: dual of #7601 with frames 3 <-> 127 swapped, main 902 = dual register 794; NEG primal 902@595 on #7601 has no clean repair; dual 794@595 repairs with K = 0, settlement rank 0/0; exact build (0 corrections, 0 premix, 1 owned observation removed); mapped back (swap, dual, COPY restore) main 902 climbs 3 -> 618 in one MOVE and its 5 owned ADDs at 595 vanish; accx strict vs #7546 ALL PASS, kappa 0.00364932638757719639692487 (above #7601); 0 lossy CLEARs; NEG dual 1289@117 has settlement rank A 0 / B >= 1 and the exact build is refused. Prints PASS_TOOLKIT_DUAL_DROP."
}
]
}
Loading
Loading